THE CENTRE FOR MARKET AND PUBLIC ORGANISATION
Parental choice of primary school in England:
what ‘type’ of school do parents choose?
Simon Burgess, Ellen Greaves, Anna Vignoles
and Deborah Wilson
November 2009
Working Paper No. 09/224
Centre for Market and Public Organisation
Bristol Institute of Public Affairs
University of Bristol
2 Priory Road
Bristol BS8 1TX
http://www.bristol.ac.uk/cmpo/
Tel: (0117) 33 10799
Fax: (0117) 33 10705
E-mail: [email protected]
The Centre for Market and Public Organisation (CMPO) is a leading research
centre, combining expertise in economics, geography and law. Our objective is to
study the intersection between the public and private sectors of the economy, and
in particular to understand the right way to organise and deliver public services.
The Centre aims to develop research, contribute to the public debate and inform
policy-making.
CMPO, now an ESRC Research Centre was established in 1998 with two large
grants from The Leverhulme Trust. In 2004 we were awarded ESRC Research
Centre status, and CMPO now combines core funding from both the ESRC and the
Trust.
ISSN 1473-625X
CMPO Working Paper Series No. 09/224
Parental choice of primary school in England:
what ‘type’ of school do parents choose?
Simon Burgess1, Ellen Greaves2, Anna Vignoles3
and Deborah Wilson4
1
CMPO and Department of Economics, University of Bristol
2
Institute for Fiscal Studies
3
Institute of Education
4
CMPO, University of Bristol
November 2009
Abstract
We investigate the central premise of the theory of markets in education, namely that parents value academic standards.
We ask what parents really want from schools and whether different types of parents have similar preferences. We
examine parents’ stated preferences and revealed preferences for schools (their actual choice of school as opposed to
what they say they value in a school). More educated and higher socio-economic status (SES) parents are more likely to
cite academic standards, whilst less educated and lower SES parents are more likely to cite proximity. More advantaged
parents choose better performing schools, particularly in areas with many schools and therefore a lot of potential school
choice. More advantaged parents also choose schools with much lower proportions of pupils eligible for free school
meals, relative to other schools available to them. Hence whilst parents do not admit to choosing schools on the basis of
their social composition, this happens in practice. Most parents get their first choice of school (94%) and this holds both
for more and less advantaged parents, though this is partially because poorer parents make more ‘realistic’, i.e. less
ambitious, choices. If, in areas where there is a lot of potential competition between schools, more advantaged families
have a higher chance of achieving their more ambitious choices that do poorer parents, this could tend to exacerbate
social segregation in our schools.
Keywords: school preferences, school choice, parental choice
JEL Classification: I20
Electronic version: www.bristol.ac.uk/cmpo/publications/papers/2009/wp224.pdf
Acknowledgements
Many thanks to Jon Johnson, Senior Database Manager at the Centre for Longitudinal Studies for providing access and
guidance for using secure data from the Millennium Cohort Study, and to Kirstine Hansen, Research Director of the
Millennium Cohort Study for her support. Thanks also to DCSF for access to the NPD dataset. We thank Jack Worth at
CMPO for providing access to his work matching infant and junior schools and participants at seminars in Bristol,
Manchester and Nottingham for comments on earlier drafts. Finally we thank the ESRC for funding this research (RES000-22-2843).
Address for Correspondence
CMPO, Bristol Institute of Public Affairs
University of Bristol
2 Priory Road
Bristol
BS8 1TX
[email protected]
[email protected]
www.bristol.ac.uk/cmpo/
CMPO is jointly funded by the Leverhulme Trust and the ESRC
1. Introduction
Since the 1988 Education Reform Act, parents in England have had the right to
express a preference for the school their child attends. This is done through an
application form to their local authority (LA), which then assigns pupils to schools1.
The Education and Inspections Act of 2006 strengthened the role of parental choice in
this assignment, banning covert selection of pupils by schools, establishing ‘choice
advisors’ for parents, and providing free transport to school for the most
disadvantaged. The reforms were based on the idea that allowing parents to choose
their school (instead of attending their local school by default) would increase
educational standards; schools would improve their performance when they were
forced to compete to attract pupils. A dual aim of the reforms was to provide fairer
access to schools2.
While the theoretical benefits of school choice as a policy are clear, the existing
evidence on its impact is mixed (see section 2). Our reading of the literature is that
competition, as it currently exists in England, has not significantly improved the
academic performance of schools. This may be because parental choice in England is
not operating as the theoretical model suggests; for example, parents may be
constrained in their choice which hinders the market mechanism. Alternatively,
parents may not value academic standards, and so schools do not compete (and raise
their standards) on this criteria. Parents may place value on the “wrong” school
characteristics (Schnieder and Buckly (2002)), such as the performance of the
schools’ sports team, or social status of the pupil intake. In this paper we investigate
this central premise of the theoretical model, namely that parents value academic
standards. Specifically we ask: do parents value academic standards as one of the
most important reasons for their choice of school, or are other characteristics more
important? What do parents really want from schools? And do different types of
parents have similar preferences across school characteristics?
We investigate these issues by first following the majority of the literature and
examining parents’ stated preferences. Second, we look at parents’ revealed
preferences for schools, i.e. their actual choice of school as opposed to what they say
they value in a school. This enables us to investigate the extent to which parents tend
to choose schools which perform better than others in their area in terms of academic
achievement compared to social and/or ethnic composition, and how this varies across
parental type. Through looking at parents’ revealed preferences we elicit information
that is unlikely to be found through analysis of stated preferences alone.
We also ask the degree to which parents’ actual choices are constrained by their
geographical location and proximity to their most preferred school3. We address this
1
Admission to an oversubscribed school is determined by the school’s admissions code. This code can
vary, but most schools operate on distance and siblings based criteria. More details can be found in
section 4, or see Pennell, West and Hind (2006).
2
In the past, access to an oversubscribed school has depended solely on proximity. Evidence suggests
that this has increased house prices in desirable catchment areas (Gibbons and Machin (2003)). This
effectively prohibits access for pupils with parents that are priced out of the market.
3
If a school is oversubscribed, the school's published admissions criteria are used to determine which
children should be offered places. These criteria have traditionally included geographical criteria such
as distance from the home to the school or the school’s 'catchment area', along with other criteria such
2
by exploring how many schools are feasible for parents and how the quality varies
within this set of schools. How many pupils attend their closest (or default) school?
And do these proportions vary by family type? Finally, are some types of families
more likely to get their first choice school?
To address these questions we use an innovative combination of survey and
administrative data that comprises the Millennium Cohort Study (MCS) and the
administrative data held on each state school pupil by the Department for Children,
Schools and Families4. The combination of rich longitudinal data in the MCS with
administrative data provides a powerful research resource which enables us to add to
the international academic literature on school choice and to the policy debate in
Britain.
The paper is set out as follows. In Section 2 we briefly review the relevant literature
both on school markets and choice. That is followed by a discussion of our data in
section 3. We then present our results and conclusions in sections 4 and 5.
2. Policy background and related literature
In the UK, the Education Reform Act of 1988 introduced a school choice mechanism
which gave parents the right to express a preference for the school their child would
attend. Funding for a particular school depended largely on the number of pupils on
the school roll, so schools had an incentive to enrol more pupils. To do this the school
had to compete with other schools in the area, along the dimensions that parents cared
about. It was thought this would provide a competitive pressure for schools to
improve their academic achievement levels. Since 1988 the market system in primary
and secondary education has continued and been strengthened with the introduction of
new types of school, more emphasis placed on parents’ right to choose their school,
and the introduction of the School Admissions Code (2007) which prohibits covert
selection of pupils by schools.
The market has clear limitations in England however. First, oversubscribed schools
cannot expand to meet demand indefinitely5. This means that although parents have
the right to express a preference for schools, their choice is not guaranteed6. Second,
failing schools often do not get shut down, which undermines competitive pressure of
the market. Rather they are ‘turned around’ with a swathe of policies aimed at
as the presence of older siblings at the school. The Schools Admissions Code (2007) standardised
schools’ admission codes.
4
The administrative data is a combination of the schools database for England, EduBase, and elements
of the Pupil Level Annual Schools Census, PLASC, along with other components of the National Pupil
Database, NPD. More details of this unique schools data set can be found in section 3.
5
Also, beyond a certain point they need official permission to expand
(http://www.dcsf.gov.uk/schoolorg/guidance.cfm?id=5 accessed 2/11/09)
6
Oversubscription criteria determine whether a parent is successful in their first choice. Most schools’
admissions are controlled and co-ordinated by their local authority (LA). In these cases, places at
oversubscribed schools are determined by criteria such as proximity and whether an older sibling
already attends the school. Voluntary Aided and Foundation schools are able to set their own
admissions criteria. Additional criteria may include whether the school was named as the ‘first
preference’ on the LA application form, and the pupil’s religious denomination in the case of faith
schools. Parents are aware of these criteria, and may decide not apply to an oversubscribed school they
deem as unfeasible.
3
improving their test results and often with additional financial investment. Finally (the
focus of this paper) it is not obvious that academic standards are necessarily of prime
concern to parents. Many other factors may be important to parents when they express
their preference for school, including, for example, the peer group and pastoral care at
the school.
A number of studies have explored the process of parental choice in detail, often
focusing on parents’ stated preferences for schools. Survey responses consistently
show that academic factors are found to be important to parents (West and Pennell
1999; Coldon and Boulton 1991; Denessen et al 2005; Kleitz et al 2000). Denessen et
al (2005) analyse the importance given to a range of reasons for choice of school in
the Netherlands. They find that ‘quality of education’ is given the most importance on
average, followed by the ‘school climate’, ‘whether the school pays attention to the
child’ and ‘order and discipline’. Factors relating to the peer group of the school are
ranked as the least important reason; the ‘possibility to come in contact with other
cultures’ was given the least importance, followed by the ‘other parents are our kind
of people’.
Kleitz et al (2000) study data generated from 1,100 charter-school parent surveys in
North Texas. They find that over 90% of parents in the survey say that education
quality in the school was important or very important to them. Class size was also
important to over 80% of parents; proximity by slightly less, while whether friends
were attending the school was reported as important or very important by the fewest
number of parents. In the US Schneider, Marschall, and Teske (2000) and Schneider
et al (1998a) find substantial race and class differences in household preferences for
schools. Parents with a college education cite diversity and teaching values as
important concerns and place less importance on test scores and discipline than
parents with less education. This may reflect the type of schools that pupils are
already in, however. Jacob and Lefgren (2007) also find evidence that parents in high
income/low income schools have different preferences. They find some support for
the notion that less educated parents have less concrete ideas about how a ‘good’
education is achieved, while more educated parents take a strong academic foundation
as given.
The dominant concern regarding school choice is that it will lead to stratification
along social or ethnic lines, as different parents value different things in schools
and/or face different constraints in realising their choice (Ball 2003; Reay, 2003;
Butler and Robson 2003). Schneider and Buckley (2002, page 134) describe the
concern that low-income and less educated parents have a “particular
susceptibility…to fall for the attraction of non-academic (and thus “wrong”) school
attributes”. Studies have generally found that parents’ stated preferences vary
according to their socio-economic background and ethnicity (Gewirtz et al 1995;
Hastings et al., 2005; Weekes-Bernard 2007; West and Pennell 1999), although Kleitz
et al (2000) report no differences between sub-groups of their population of
‘choosers’.
Interpreting this evidence can be problematic as parents’ stated preferences may differ
from their true preferences. This may be particularly true regarding socially sensitive
issues such as school ethnic composition (Jacob and Lefgren, 2007). Academic
standards may only be mentioned by parents as a ‘socially desirable’ response, for
4
example, when school composition is in fact more important to them. Unless faced
with choices in a realistic setting, revealed preferences are not elicited.
Schneider and Buckley (2002) report the “remarkable consistency” in the verbal
(survey) responses of parents about what they value in schools. They cite six papers
based on survey data that find that academic and teacher quality are the most
important reasons given by parents in their choice of school. They note, however, that
the “optimistic” conclusions based on survey responses are in contrast to evidence
from revealed preferences. They use an online schools database in Washington, DC to
monitor parent search behaviour as an indicator of parent preferences. Through
analysing this search behaviour, they find a “strong bias toward accessing the
demographic characteristics of the student population, which is in marked contrast to
verbal reports about the importance of race” (Schneider and Buckley 2002, page 138).
The authors also cite other work in the US which supports this finding: Weiher and
Tedin (2002) show that when choosing charter schools parents in Texas were likely to
“sort themselves along racial/ethnic lines . . . in spite of their expressed preferences,
rather than in conformity with those expressed preferences” (Weiher and Tedin
(2002), page 91, italics added).
Finally, Buckley et al (2006) use survey data in Chile to assess whether stated and
revealed preferences are consistent. They find that while all parents in Santiago say
they are seeking strong academic programs in their children’s schools, they actually
choose schools that are widely different on academic quality but similar on
socioeconomic dimensions. The authors conclude that parents choose “class – not the
classroom”, which stated preferences do not reveal. Our data allow us to investigate
both stated and revealed preferences for primary schools in England and to compare
the two.
3. Data
Our primary data source is the Millennium Cohort Study (MCS), which is a
longitudinal dataset funded by the Economic and Social Research Council (ESRC)
and a consortium of Government Departments headed by the Office for National
Statistics (ONS). The MCS was sampled from all live births in the UK, which for
England and Wales took place over a 12 month period from 1 September 2000 and 31
August 2001. This period coincides with the school academic year, meaning that
children in the study in England and Wales form an academic cohort. The MCS is
designed to chart the initial health, social and economic conditions of new births in
the new millennium (MSC Guide to the Datasets 2008).
For England, the sample was selected from a random sample of electoral wards,
disproportionately stratified to ensure adequate representation of deprived areas,
defined as the poorest 25% of wards based on the Child Poverty Index (CPI)7, and
also areas with high concentrations of Black and Asian families8. Weights provided
with the data are applied to the analyses reported here to correct for the over7
The CPI is defined as the percentage of children under 16 in an electoral ward living in families that
were, in 1998, receiving at least one of the following benefits: Income Support; Jobseekers Allowance;
Family Credit; Disability Working Allowance, and is therefore a proxy for neighbourhood deprivation.
8
‘High’ is defined as above 30% of the ward. Proportions are based on the 1991 national census.
5
representation of disadvantaged and high ethnic minority wards and so to enable us to
draw nationally representative inferences. The sample draws from all children that are
living in the UK at age nine months and eligible to receive Child Benefit at that age.
Child Benefit is a payment made to all main carers and so has extensive coverage; it
excludes only those whose residency is temporary9.
Our study uses data from England only, as school policy and admission arrangements
differ between countries in the UK and because the administrative data for schools we
use is collected in England only. 5,748 children were sampled from ‘advantaged’
wards, 5,946 from the ‘disadvantaged’ and 3,461 from the wards with a proportion of
ethnic minorities greater than 30%. Due to ineligibility of some families and refusal to
participate by others, the final sample achieved was 11,533, with a breakdown of
4,617, 4,522 and 2,394 for the respective groups.
Our paper uses the third survey of the MCS, carried out in 2006 when children in the
study are five years old. The timing of the third survey was deliberate: the children
were beginning primary school and a main objective of the MCS was “to record the
child’s transition to primary school, from the point of view of their parents and their
school” (MCS Guide to the Datasets 2008; page 13). A consultation process in 2004
decided that it would be useful to assess issues around school choice, including
detailed questions on parents’ preferences for schools and information on the schools
they actually applied to (MCS Consultation Group 2004). Variables such as the
reasons parents give for their ‘first choice’ of school, the most important reason in this
choice, and whether parents achieved their desired school are taken from this section.
The data allows us to investigate parents’ stated preferences for schools and to
compare the findings from the MCS with previous national and cross-country
evidence.
We also explore parents’ revealed preferences for schools. This involves comparing
schools in the surrounding area with the school named as the ‘first preference’ on the
Local Authority (LA) application form. We compare the first preference school with
other schools in the locality in relation to a number of characteristics, which requires a
detailed dataset containing information on all primary schools in England. To create
this dataset we combine variables from two data sources; EduBase and the National
Pupil Database (NPD).
EduBase is a register of all educational establishments in England and Wales,
maintained by the Department for Children, Schools and Families (DCSF). EduBase
contains administrative information for each school, whether it is in the state or
independent sector. This includes the type of school (whether voluntary aided,
voluntary controlled etc.), phase of education (primary, middle or secondary) and its
exact location (postcode).
The NPD is an administrative dataset that covers all pupils in state schools in
England. This dataset contains information on pupils’ attainment at each Key Stage
level of the National Curriculum, and also information from the pupil level census
such as each pupil’s eligibility for free school meals, recognition of any special
9
The Department for Work and Pensions (DWP) estimates that 97.2% of children received Child
Benefit by 7 months of age in 2000.
6
educational need and whether they have English as an additional language. Providing
this information for the census (previously known as PLASC, now the Annual
Schools Census) is a statutory requirement for all maintained schools in England; data
should therefore be accurate and reliable. We aggregate the pupil level data to school
level, which gives us school level variables such as the percentage of pupils in the
school that are eligible for free school meals (FSM), a proxy for low income; the
percentage that have any special educational need (SEN); and the percentage that
have English as an additional language (EAL). From the attainment data we also
construct a measure of each school’s average Key Stage 2 (KS2) points score, and the
percentage of pupils that achieve level 5 in all subjects at KS210. These indicators are
calculated for the cohort of pupils taking the tests in 2003, which were published in
school performance tables in 2004 and so were available to MCS respondents when
making their school choices11.
Variables taken from EduBase and those constructed from the NPD are combined to
form a detailed schools dataset. This dataset does not include information on average
attainment or composition for schools in the independent sector as information for
these schools is not collected by the DCSF. Our schools dataset enables us to compare
schools in the local area with the schools nominated by parents on their LA
application form. In section 4 we explain how we construct an indicator for whether
any particular school represents a feasible choice for parents in terms of the distance
between the home and the school and the school’s catchment area.
4. Results
4.1 Stated preferences for schools
The MCS survey contains a detailed set of questions on parents’ preferences for
schools, their actual nominations on their LA application form, whether they got the
school that they named as ‘first preference’ on that form, and whether they are happy
with their child’s allocated school12.
The main respondent (usually a parent) is first asked whether they applied for a
primary school place through their local authority’s application form. This is the
standard method of application: all parents who would like a place for their child at a
state primary school should complete the form. Parents are then asked to give details
(name, address and postcode) of the schools they put as their first, second, and third
nominated preferences on the application form if applicable.
Figure 1 shows the percentage of parents that either expressed no preference on their
LA application form, or one, two or three preferences. Around a quarter say they did
not complete the LA application form for schools. We believe this proportion is not
10
Level 5 is above the expected level of attainment at KS2 set by government. This indicator therefore
represents the proportion of pupils that achieve highly at the end of primary school.
11
School performance tables are available online each year from the Department of Children Schools
and Families (DCSF), for example see: http://www.dcsf.gov.uk/performancetables/primary_08.shtml,
where a variety of statistics on school performance are published.
12
Full details of the questions asked at wave 3 of the survey can be found at
http://www.cls.ioe.ac.uk/studies.asp?section=0001000200010011.
7
representative; statistics collected from a sample of LAs suggest that over 90% of
parents complete this form as it is compulsory for allocation to state primary schools.
This would seem to imply that our data suffer from one or more of: potential recall
bias, whereby a proportion of parents cannot recall, even a short time afterwards
(around one year), that they completed the LA form; parents misunderstanding the
question in the survey; parents wanting to take a short-cut in the questionnaire.
Following this question, parents are asked what factors were important to them when
they considered which school to nominate as their first preference. Specifically, they
were presented with a list of reasons for their school nominations and asked “Which
of these factors were important in choosing to put this school as first choice on the
form?” More than one reason could be given. Most parents give multiple reasons for
choosing a particular school (the mean number of reasons given was 4.09), although
this varies slightly by parent characteristics such as education and socio-economic
status.
Table 1 shows the percentage of main respondents that named each reason on the list
as an important factor in their first choice nomination. 67.28% of parents say that
proximity to the family home and 62.87% say that a ‘general good impression’ of the
school is important. These two reasons are the most commonly cited but whether
siblings attend the school and academic standards are also mentioned by a large
percentage of parents (46.93% and 43.55% respectively). Small class sizes are
mentioned by around a fifth of parents (18.70%) and religion by around a tenth
(10.24%). Only 3.91% of parents state that the ethnic mix of the school is an
important consideration, but 22.43% mention that where the child’s friends are likely
to apply is an important factor. This final consideration may mean that parents choose
schools that have a pupil intake with similar characteristics, such as ethnicity and
social class.
As shown in Table 1, parents give a large number of reasons for their first choice
school. For ease of analysis we group these reasons into ten categories, as shown in
Table 2. Appendix 1 gives the details for the grouping. The majority of parents
(67.95%) cite proximity and/or ease of travel as important reasons for their naming a
school as their first preference on the LA application form. A high percentage gives a
‘general good impression’ of the school (62.90%), as does the school’s characteristics
and facilities (52.36%). Whether a sibling attends the school is another important
reason often given (46.93% of parents), with academic standards slightly less
(43.55%). The reasons given for the choice of first choice school in our data are
consistent with previous evidence on parents’ stated preferences for schools as
discussed above.
In our data the majority of respondents give more than one stated reason as important
in their decision for their first nominated school. Although we do not know the
ranking for these stated reasons, we do know the reason they recall was the ‘most
important’. Respondents are asked: “And, which of these was the single most
important reason (for putting this school as first choice)?” From this question we can
identify the main reason for their choice, although caveats apply. These are stated
preferences, and when answering this question parents may not consider all reasons,
may be reluctant to give socially undesirable responses, or may have forgotten initial
decisions which later conditioned their choices, for example moving house. There
8
may also be a certain amount of recall bias, or parents justifying their decision ex
post. These shortcomings are discussed more fully below and in a companion paper to
the present one (Burgess et al. 2009).
As before we group the list of most important reasons into ten categories. The
percentage of parents giving each grouped reason as the ‘most important’ is given in
Table 3. Around a quarter of parents say that proximity and/or ease of travel are the
most important reasons for their school choice, with a similar proportion saying that a
sibling already attending the school is most important. Around 16% of parents say
that academic standards are the most important consideration, and about the same
proportion say a ‘general good impression’. Very low proportions of parents cite other
reasons as the ‘most important’ in their choice of first preference school on the LA
form. The responses for ‘most important’ reason for choice of school are consistent
with previous evidence on parents’ stated preferences for schools.
Given the concern that differences in parents’ preferences may lead to social
stratification in schools, we also investigate the extent to which parents’ stated
preferences vary by socio-economic status and ethnicity (see also Henig 1996; Lee at
el 1996).
Table 4 shows the variation in ‘most important’ reason given, by the main
respondent’s highest educational or vocational qualification. 40% of those with no
educational or vocational qualifications say that proximity is the most important
factor in their choice. Those with no qualifications are almost twice as likely to give
proximity as their most important reason as compared to those with at least a degree
or equivalent qualification. In contrast, those with degrees are over twice as likely to
say that academic standards are the most important reason: 20.27% of those with at
least a degree give this response, compared with 7.80% of those with no
qualifications. A similar pattern exists for the percentages giving a ‘general good
impression’ as the most important reason. More educated individuals are also over
twice as likely to give religious grounds as the most important reason.
Table 5 shows that a similar pattern exists between those with low and high levels of
socio-economic status.13 Around one third of those in the lowest SES quintile gave
proximity as the most important factor, compared to fewer than one in five of those in
the highest quintile citing this reason. Higher SES parents are more likely to cite
academic standards and a ‘general good impression’ of the school as the most
important reason. As with variation by education, those with higher SES are also
more likely to give religion as the most important reason for their school choice.
The evidence discussed in section 2, however, suggests that it is important to analyse
parents’ revealed preferences for schools rather than relying on survey data or stated
preferences alone. In addition to removing any bias caused through a desire to
conform to socially desirable norms, parents may also conflate preferences for
13
Our measure for socio-economic status is derived from principal component analysis and includes
the following parental variables: whether the main respondent reported financial difficulties in wave 3
of the survey, the highest NS-SEC (National Statistics Socio-economic Classification) of the
household, housing tenure, and logged equivalised income. Those in the lowest quintile have the lowest
level of socio-economic status, and those in the highest quintile have the highest. See appendix 2 for
full details of the procedure.
9
schools. For example, parents may state that proximity is the most important factor in
their school choice when in fact they have moved house to be close to a school that is
their most preferred for other reasons. Through using revealed preferences, i.e.
looking at the characteristics of the school they actually nominate as their first
preference, we can identify parents’ preferences for schools with more certainty than
by using survey data alone.
4.2 Revealed preferences for schools
To understand the extent to which our data reveal parental preferences we need to
briefly recap the primary school admission process in England. Parents are invited to
apply for a school place through the local authority (LA) application form, on which
up to three schools can be named as preferred schools14. They are named in order of
preference, so the school named first on the application form should be the school that
parents most prefer (for qualifications to this see below). Using this form the LA
allocates pupils to schools, aiming to maximise the number of parents getting a place
at one of their preferred schools. When a school is oversubscribed published
admissions criteria determine which pupils will be admitted. Admissions criteria for
community and voluntary controlled schools are decided by the local authority and so
are common across schools in any one local authority. Voluntary aided and
foundation schools control their own admissions and so criteria may vary.
Applications to all schools, no matter what type, must be through the co-ordinated
admissions form. The allocation process is as follows:
1) Parents complete and return their application forms to the LA, stating their
preferences in order.
2) All schools nominated on the application form are asked whether a place for
the pupil is available at the school. At the time of application by the MCS
cohort, whether the parent named the school as 1st, 2nd or 3rd preference on the
LA form was used as admissions criteria by schools in some LAs. In others,
the rank order was not taken into account15.
3) All schools have a binding maximum capacity for entry places. Whether the
school has a place for the pupil depends on the number of entry places it has in
a given year. If a school is oversubscribed, entry is usually determined on
proximity to the school, or ‘catchment area’ after siblings, pupils in care and
those with a statement of special educational needs have been accepted.
4) If a pupil is deemed to be eligible for a place in more than one school, the
allocation will be determined by the LA through considering which school has
been ranked highest on the parent’s application form.
14
A few LAs invite more preferences, but the mode is 3. See Coldron et al (2008) for more details.
Under ‘equal preferences’ admission criteria, where rank order is irrelevant, parents’ incentives are
consistent with naming their most preferred school as the highest rank on the LA form. Under ‘first
preferences first’, parents may be more strategic. The admission criteria of ‘first preferences first’
means that admissions bodies prioritise pupils who named the school as first choice on the preference
form. This creates a big risk for parents, as naming an oversubscribed school as your first preference
means you reduce the likelihood of entry into your ‘default’ school, where it is different. Coldron et al
(2008) surveyed LAs and found that around 32% used ‘first preference first’ admission criteria in
2006/2007, two years after after our cohort applied to primary school. In addition to this, schools that
control their own admissions may apply ‘first preference first’ criteria even where the LA does not.
15
10
All parents intending to send their child to a state primary school should complete the
LA application form. If parents say that they applied for a school place through the
LA application form, the MCS survey then asks “What is the full name of the school
that you put as first choice on the form?” as well as the address and postcode of the
school. This question is repeated for the second and third choices if made. We take
the school named as the first preference as the ‘revealed’ preference of the parents.
We are aware that the revealed ‘choice’ (first preference) we observe may be
constrained in some way. The school that parents name as the first choice on the
application form may reflect resigned acceptance rather than true preference if parents
view their truly desired first preference as unfeasible. Expressing a preference for the
most oversubscribed school in the neighbourhood, for example, especially if it is far
away, may seem like a wasted choice. This may be particularly true for families living
in disadvantaged neighbourhoods who are likely to be further from the most
oversubscribed and/or high attaining schools. In areas which operated under ‘first
preference first’ admission criteria in 2004/2005, the risk of putting an oversubscribed
school as the first preference is amplified (see footnote 15). To at least partly address
this problem we incorporate a further question in the MCS in our analysis. Parents
were also asked “were there any schools you would have liked [cohort child's name]
to attend but chose not to apply to?” and were then asked to name this school. We
represent parents’ revealed preferences for primary school by parents’ first nominated
school if they do not say there was another school they would have liked to apply to
but didn’t, and the school that parents would have liked to have applied to but didn’t
if this is applicable16.
In practice when filling in the LA application form, parents are choosing between a
range of ‘local’ schools, often within their own local authority. In the context of
parents’ revealed preferences for schools we want to look at the choice of school out
of all such schools that are feasible for the parent, which we call the parent’s ‘feasible
choice set’. By comparing the characteristics of the chosen school with those in the
choice set, we can elicit parents’ revealed preferences across these characteristics.
Given that all schools in the choice set are feasible, the school the parents name on
their application form should be the parents’ most truly preferred school from the set
of schools they can reasonably access. We can then compare this revealed preference
to our evidence of parents’ stated preferences.
4.3 Defining the ‘feasible choice set’
The process of defining this feasible choice set is essential. At first thought it may
seem that all schools within a certain distance are feasible for parents, for example
within 3km from the home. This distance may vary between rural and urban areas due
to different norms and transport links, but the same concept would hold in all areas:
within a certain travel distance all schools are considered feasible by parents. By
looking at school choice within this set of schools we could then identify parents’
revealed preferences. There is a flaw with this line of enquiry however; parents will
be unlikely to waste a nomination on a school they know is unattainable. Within any
specified distance boundary, some schools will be popular and have a small
16
663 parents, or around 7% of the sample say that they would have liked to apply to another school
but didn’t for some reason. 30.77% of these parents say they did not apply because they were doubtful
the child would get in, 30.17% because the school was too far away or because of travel constraints.
11
catchment area. Other schools will be less popular and may admit pupils from a larger
distance. We assume that parents have some knowledge of how popular each school
is, or where pupils generally travel from to get to the school. Popular schools with
small catchment areas may not be a realistic choice, and so should not be included in
the parent’s feasible choice set. We incorporate this into how we calculate our feasible
choice set by first defining each school’s catchment area as follows:
1) Calculate the straight-line distance between the school and home address for
all pupils in the reception year group in 2004. This is the entry cohort in the
year before the MCS cohort entered primary school and therefore the year the
MCS respondents applied to schools.
2) Calculate the 80th percentile: the straight line distance within which 80% of
the pupils in the previous year’s intake live.
We calculate this catchment area based on the previous year’s entry as this is the year
that is most relevant to prospective MCS parents. Most LAs provide information on
the number of applications for entry and the number of places possible to be allocated
in the previous year, which gives an indication of relative supply and demand. This is
presented alongside other school information in the application booklet and so is
easily accessible. Some LAs also give further information on oversubscribed schools,
for example how far away the furthest pupil lived from the school in the previous
year. To be in the feasible choice set, we also impose that the school must be in the
same LA and within 20km of the pupil17.
So, to summarise, for every MCS respondent in our sample we calculate the set of
schools that are feasible for them to choose between. To be in this feasible choice set
a school has to be in the same LA and within 20km of the family’s home address, and
the family must live within the radius from which 80% of the school’s intake has been
drawn the previous year.
Parents can change or manipulate their feasible choice set in various ways. Parents
may decide to move house to improve the likelihood of entry into their desired school;
they may invest more time in religious activities to gain entry in faith schools; they
may opt out of the state education system altogether and pay for an independent
school. We abstract from these issues here, but in a companion paper we restrict the
sample of parents to those who have not moved house since the cohort member was
born (Burgess et al. 2009); the results from this sample and the full sample of families
are not substantively different. We exclude those in independent schools in our
sample as we look at preferences expressed on LA application forms18.
17
This final restriction is simply to ensure that schools with very large catchment areas, such as
boarding schools, are excluded from the feasible choice set.
18
In practice, 89 parents apply to the local education authority for a state school place and place their
child in a fee paying school. This represents 22% of all pupils in fee paying schools. 51 of the parents
who applied to both the state and independent sector were not allocated their first nominated school in
the state sector, suggesting some ex post decision making. We abstract from this issue because of the
small sample size of the subgroup.
12
4.4 The ‘type’ of schools available in the feasible choice set
Table 6 shows that schools in the average feasible choice set vary across different
types of family19. Schools in the feasible choice set for parents in the lowest SES
quintile have a much higher average FSM rate (22%) compared to those in the choice
sets within which families in the highest SES quintile are located (11%). Parents in
the lowest SES quintile have schools which have higher proportions of pupils with
EAL than parents in the highest SES quintile (12% compared to 8%), and pupils are
also more likely to have any level of SEN (21% compared to 17%). Schools available
for low SES parents are less ‘white’, and more likely to have lower proportions of
high attaining students; an average of 17% of pupils get all level 5 at Key Stage 2 for
low SES parents, compared with 23% for high SES parents. High SES parents are
more likely to have access to voluntary controlled schools. 11% of schools for high
SES parents are voluntary controlled, compared with 6% of schools for low SES
parents. Similar differences in the characteristics of available schools exist between
those with no educational or vocational qualification and those with at least a degree.
These differences in available schools across parent types highlight why it is
important to consider a parent’s school choice relative to all others available in their
feasible choice set. Without considering choices of school in a relative context we
would estimate a parent’s preferences based on absolute attainment of schools in their
area. Table 6 shows that on average, all types of parents have some variation across
school characteristics within their feasible choice set, which suggests most parents
have some choice between school types. They could, for example, feasibly apply to a
high scoring school with a low proportion of pupils eligible for FSM or a low scoring
school with a high proportion of pupils eligible FSM (both defined relative to their
local area20). We argue that the availability of this choice to most parents can enable
us to genuinely illicit parents’ revealed preferences.
4.5 Parents’ revealed preferences: the characteristics of their chosen school
Given the definition of the feasible choice set above, in this section we ask: what type
of school do parents choose (i.e. put first on the LA form), relative to all the others
available to them? Do they choose the most highly attaining school in terms of KS2
results, the school with the lowest proportion of low income students, or the nearest
school? To look at these questions we rank the schools in each pupil’s feasible choice
set by each of these three school characteristics in turn. A school ranked number 1
will be the school with the highest KS2 mean score, the lowest proportion of pupils
eligible for FSM, or the one to which they live closest to in the feasible choice set,
depending on which school characteristic they are ordered by21.
19
Different parent types also have different numbers of schools in their feasible choice set, and we
account for this below.
20
A “low scoring” school is defined as a school with KS2 attainment (averaged over 3 years) below
that of the median school in the pupil’s feasible choice set. A “high scoring” school is defined as a
school with KS2 attainment (averaged over 3 years) above the median school. If the school has the
same average KS2 attainment as the median school, they are randomly assigned to either the “high” or
“low" scoring category. The same methodology is used to define “rich” and “poor” schools relative to
all schools in the feasible choice set.
21
We could choose to rank these characteristics either from low to high or high to low. By using the
ranking here we do not mean to imply that schools with low numbers of FSM-eligible students are in
13
Table 7 shows the rank of school that parents choose when schools are ranked by their
KS2 mean. The table splits parents by their SES quintile and reveals differences in
school choice across parent type. A rank of 1 means that, on average, the parent SES
type chooses the most highly performing school in their feasible choice set, a rank of
2 means they choose the second most highly performing on average, and so on.
Different parent types have different numbers of schools available in their feasible
choice sets. Those in the lowest SES quintile have 5.46 schools in their feasible
choice set on average, compared with 5.28 for those in the highest SES quintile.
Although this is not a large difference, to control for the number of schools in the
feasible choice set we present the statistics separately by the number of schools
available. Column 1 of Table 7 shows the rank of school chosen for parents that have
5 schools in their feasible choice set, column 2 for those that have 8, and column 3 for
those that have 10. Rows show the rank of school chosen for parents of different
levels of socio-economic status (SES). This enables us to compare the choices across
parent types while holding constant the number of schools they are able to choose
between. Sample sizes for each group are also given.
Consider first column 1, comparing parents who have five schools in their feasible
choice set. On average, parents with low SES choose a lower ranked, i.e. lower
performing, school, than those in the highest SES quintile: 3.45 compared to 2.69
(both out of five available; a difference in ranking of 0.76). Table 7 shows that there is
more variation between parents of high and low SES in areas where there are more
schools in the feasible choice set. The difference in ranking is 1.65 for those with
eight schools, and 2.54 for those with ten22.
Table 8 shows that parents’ preferences are more varied when comparing school
choices over schools ranked by the proportion of students that have free school meals
(FSM). For those with eight schools in their feasible choice set, the average parent in
the lowest SES quintile chooses a school that is ranked 5.26 out of 8. The average
parent in the highest SES quintile chooses a school that is ranked 3.72 (again out of 8)
on average, 1.54 schools higher in the ranking. For those with ten schools in their
feasible choice set, the difference between the highest and lowest SES is around 4
schools in the ranking (8.12 compared with 3.82), although the sample size is small in
this case.
This analysis shows that parents of different socio-economic status choose a school
based on different characteristics. Those with high SES tend to choose a school
ranked more highly in its academic quality, and with a lower proportion of pupils
eligible for FSM. We are aware that any interpretation of these differences must be
cautious however, as they could arise for a number of reasons:
1) The high performing school has covert admission requirements. This could
mean that parents’ preferences are the same, but some parent types are more
any way ‘better’ than those with high numbers. Rather, we are illustrating the process of parental
choice between schools with different proportions of different student types.
22
The difference is also clear when accounting for the total number of schools in the feasible choice
set. Using the ratio of the school rank chosen to the total number of schools, the differences between
high and low SES are 0.152, 0.20 and 0.25 for the 5, 8 and 10 school cases respectively.
14
constrained than others. We have accounted for this as fully as possible in our
definition of feasible choice sets, as pupils in the previous year travelled at
least as far as pupil i. We cannot, however, account for the extent to which
knowledge of such covert selection practices may stop some parents from
applying to such schools.
2) Within the feasible choice set, the high performing school is still further away
for low income families.
3) All parents attend the local school. This means that in disadvantaged
neighbourhoods it looks like parents ‘choose’ high-FSM schools explicitly,
and the opposite in advantaged neighbourhoods.
4) Parents have different preferences for schools. In rural areas and areas with
fewer schools, we do not observe these different preferences as parents must
choose the same schools. In urban areas and areas with many schools in the
feasible choice set (so there are more options) parents are free to choose and
we observe the differences in preferences more clearly.
Our data does not allow us to investigate point 1. This would involve qualitative
research into schools’ admission requirements (in practice, not simply as stated on
admissions forms). For some recent evidence into secondary school admissions see
Pennell et al (2006) for London and Coldron et al (2008) for England. Points 2 and 3
are related. If all parents nominate the closest school, either through actual preference
or constraints on their choice, then our model of revealed preferences will simply
show that parents prefer schools in their immediate neighbourhoods, no matter what
their characteristics. If pupils with FSM are clustered in areas where other pupils also
have FSM, then the school their parents choose will have a high proportion of FSM
pupils. Similarly, if pupils are clustered with similar ability types, then high ability
pupils will appear to ‘choose’ high attaining schools, and low ability pupils will
appear to choose low performing schools.
Table 9 gives some evidence that parents in all SES quintiles attend a school close to
their home. For those with five schools in their feasible choice set, those with low
SES nominate a school which is ranked 2.47 on average. This is between the 2nd and
3rd closest school in the feasible choice set and similar to high SES parents who
nominate a school which is ranked 2.10 on average. For those with eight schools in
their feasible choice set parents in all SES quintiles choose a school which is between
3 and 3.5 schools away on average, so again there is little variation. There is slightly
more variation for those with 10 schools, between 3.53 for the lowest SES quintile
and 5.13 for the fourth SES quintile.
In each size of feasible choice set, parents choose a close school. Relative to other
school characteristics, there is remarkably little variation between SES quintiles: all
parent types choose a close school. This suggests that either parents choose a close
school, no matter what its characteristics, or that parents self select into
neighbourhoods with their desired school type. If neighbourhood characteristics are
correlated with school characteristics and parents choose the closest school, this
would explain the variation in school rank chosen when schools are ranked by school
performance and the proportion of pupils with FSM.
15
4.6 Who gets their first choice?
The average parent has five schools with varying characteristics within their feasible
choice set, which suggests that parents do have some degree of school choice. What
about the outcome of this choice process? In our data a very high proportion of the
parents who completed the LA form report getting their first choice of school (94%).
This is consistent with individual reports collected from LAs23. Furthermore, we find
little variation in the proportion of parents getting their first choice school across SES
quintiles, main respondent’s education level, or IMD deciles24. It may be however,
that parents of disadvantaged backgrounds are adjusting their preferences on the LA
form to reflect a resigned choice and we investigate this further.
Figure 2 shows that there is variation in the proportion of families getting their highest
nominated school between different types of geographical areas. Those with fewer
schools in their local area are more likely to get a place in their most highly
nominated school, with the exception of those with no schools within 3km. Those in
more densely populated areas are less likely to get their most highly nominated
school. This suggests that geographical location and population density are significant
factors in whether a pupil is allocated their most highly nominated school. This may
be because there are more excess school places in areas where population density is
lower, because parents are willing to travel further, or are willing to apply to a risky
‘ambitious’ school in more densely populated areas.
Taking this further, we find that although the proportion of parents getting their first
choice of school does not vary by SES, the price of being ambitious in your school
choice does. In the sample as a whole, parents that apply to the highest ranked school
in their feasible choice set have a lower probability of admittance. For example, Table
10 shows that 91% of parents that choose the most highly ranked school in terms of
mean KS2 score are admitted compared to 97% of parents that choose the least highly
ranked school. Thus there is a potential risk or penalty associated with choosing
higher ranking schools. This risk is more evident for students in the lowest SES
quintile. Only 80% of low SES parents that choose the school ranked most highly in
terms of academic achievement are admitted, compared with 91% of those in the
highest SES quintile. Although on average parents of different types are equally likely
to gain admittance to their ‘most preferred’ school on the LA application form, this is
not the case if parents of low SES apply to a school with the best academic
achievement in their area. This is consistent with qualitative evidence from WeekesBernard (2007) who surveyed British Black and minority ethnic parents, and found
23
We contacted a small number of LAs to find out the proportion of pupils that were allocated to their
first choice school. The high proportion is also consistent with data for secondary schools, which is
collected by the government. In 2008 82% of families got their highest preference school, ranging from
64.5% in London to 92% in the North East.
http://www.dcsf.gov.uk/rsgateway/DB/STA/t000791/index.shtml
24
IMD is the Index of Multiple Deprivation. This index combines a number of indicators, chosen to
cover a range of economic, social and housing issues, into a single deprivation score for each small
area in England. This allows each area to be ranked relative to others according to their level of
deprivation
(http://www.communities.gov.uk/communities/neighbourhoodrenewal/deprivation/deprivation07/) .
The Indices of Deprivation 2007 have been produced at Lower Super Output Area level (of which there
are 32,482 in the country).
16
that many felt constrained in their choice of school, ruling out schools that were
deemed unattainable.
5. Conclusion
School choice is central to the English system of school admissions, and has been a
major element of government policy for more than two decades. Parental school
choice has its supporters and its critics. Those who believe in market power suggest
that giving parents the right to choose the school their child attends will tend to cause
schools to compete with one another and raise academic standards as a consequence.
This positive view is implicitly based on the assumption that parents will choose
schools that are more effective in raising levels of academic achievement. Critics
argue that parents may not, however, choose on the basis of academic achievement
but rather on other potentially divisive factors, such as ethnic and/or social class
composition. We first investigate these issues by following the majority of the
literature and examining parents’ stated preferences. We then use our preferred
strategy and look at parents’ revealed preferences for schools based on the choice they
make on their primary school admission form, comparing the characteristics of the
schools they choose relative to all other feasible schools available to them. This
method is an improvement on relying on stated preferences alone, since what parents
feel comfortable admitting as the reasons for their school choice may not match their
real motivations, as revealed by their actual choice of school.
Around a quarter of parents cite proximity and ease of travel as the most important
criteria behind their choice. By contrast, only 4% of parents state that the ethnic mix
of the school is an important consideration. On the face of it therefore, parents do
choose, if not on academic standards per se (only half of parents cite this as being
important and just 16% put it as the most important factor), but on the convenience
and characteristics of the school. We do find however, that the reasons cited for
school choice vary across different types of parent. More educated and higher SES
parents are more likely to cite academic standards and a “general good impression” of
the school, whilst lower educated and lower SES parents are more likely to cite
proximity.
When we turn to revealed preference we are careful to compare the parents’ actual
choice of school, i.e. the school they put first on their local authority application form,
with other feasible schools in their neighbourhood25. This is important as,
unsurprisingly, different types of schools are available for higher and lower SES
parents and their children. On average, more advantaged parents choose a better
performing school relative to others in their choice set, particularly in areas with many
schools and therefore a lot of potential school choice. More advantaged parents are
also likely to choose schools with much lower proportions of pupils eligible for free
school meals relative to other schools available to them. Whilst parents do not admit
to choosing schools on the basis of their social composition, there is evidence that this
is happening in practice. Conversely, lower SES parents are less likely to apply to
socially advantaged schools, perhaps because they feel they will not be able to get
25
In a companion paper to this (Burgess et al 2009) we further investigate the extent to which parents’
first preferences on their application form represent their true preference as opposed to constraints they
face in the current school choice system.
17
into such schools (perhaps due to unobservable illicit admission requirements or,
equally, social norms). These results may not however, be purely driven by different
types of parent having different preferences for a particular type of school. When we
undertake formal modelling of the choice process (Burgess et al. 2009), we find that
in fact many apparent differences in parental preference are insignificant. What
matters more is location. We find strong and compelling evidence that parents do tend
to choose a nearby school. This of course would also tend to undermine the process of
school choice which relies on parents choosing better schools rather than closer ones.
Is there any truth then to the idea that poorer children are de facto excluded from
some higher performing or socially advantaged schools? In general most parents get
their first choice of school (94%) and this is equally true for both more and less
advantaged parents. Our results indicate however that this is partially because poorer
parents make more ‘realistic’, i.e. less ambitious, choices on their LA application
form, reflecting differences in preference and/or constraint (Burgess et al 2009).
Where low SES parents make ambitious school choices they have a much lower
chance of securing their first choice of school than similarly ambitious high SES
parents. Geographical location and population density are also significant factors in
whether a pupil is allocated their most highly nominated school. Again this has
implications for school choice policy. If, in areas where there is a lot of potential
competition between schools, more advantaged families have a higher chance of
achieving their more ambitious choices than do poorer parents, this could tend to
exacerbate social segregation in our schools.
18
References
Ball, S. J. (2003) Class strategies and the education market: The middle classes and
social advantage. London: Routledge Falmer.
Burgess, S., Greaves, E., Vignoles, A. & Wilson, D. (2009) What parents want:
School preferences and school choice. CMPO working paper no. 09/222.
Butler, T. & Robson, G. (2003) Plotting the middle classes: Gentrification and circuits
of education in london. Housing Studies, 18(1): 5-28.
Coldron, J. & Boulton, P. (1991) ’happiness’ as a criterion of parents’ choice of
school. Journal of Education Policy, 6(2): 169-178.
Coldron, J., Tanner, E., Finch, S., Shipton, L., Wolstenholme, C., Willis, B., Demack,
S. & Stiell, B. (2008) Secondary school admissions. Research Report No
DCSF_RR020.
Denessen, E., Driessena, G. & Sleegers, P. (2005) Segregation by choice? A study of
group-specific reasons for school choice. Journal of Education Policy, 20(3):
347-368.
Elacqua, G., Schneider, M. & Buckley, J. (2006) School choice in chile: Is it class or
the classroom? Journal of Policy Analysis and Management, 25(3): 577-601.
Gewirtz, S., Ball, S. J. & Bowe, R. (1995) Markets, choice and equity in education.
Buckingham, Open University Press.
Hastings, J., Kane, T. J. & Staiger, D. (2005) Parental preferences and school
competition: Evidence from a public school choice program, (SSRN).
Henig, J. (1996) The local dynamics of choice: Ethnic preferences and institutional
responses. . In B. Fuller, R. F. Elmore and G. Orfield (Eds.), Who chooses?
Who loses? Culture institutions and the unequal effects of school choice. New
York: Teachers College Press. .
Jacob, B. A. & Lefgren, L. (2007) What do parents value in education? An empirical
investigation of parents' revealed preferences for teachers. Quarterly Journal
of Economics, 122(4): 1603-1637.
Kleitz, B., Weiher, G. R., Tedin, K. & Matland, R. (2000) Choice, charter schools,
and household preferences. Social science quarterly, 81(3): 846-854.
Lee, V. E., Croninger, R. G. & Smith, J. B. (1996) Equity and choice in detroit. . In B.
Fuller & R.F. Elmore (Eds.), Who chooses, who loses? Culture, institutions,
and the unequal effects of school choice (pp. 70-94). New York: Teachers'
College Press.
MCS Consultation Group (2004)
http://www.cls.ioe.ac.uk/studies.asp?section=0001000200010011 accessed 1/11/09.
MCS guide to the datasets (2008)
http://www.cls.ioe.ac.uk/studies.asp?section=0001000200010015 accessed 2/11/09.
Pennell, H., West, A. & Hind, A. (2006) Secondary school admissions in london.
Centre for Educational Research; Department of Social Policy; London
School of Economics and Political Science, Clare Market Papers No. 19
Reay, D. & Lucey, H. (2003) The limits of ‘choice’: Children and inner city
schooling. Sociology, 37(1): 121-142.
Schneider, M. & Buckley, J. (2002) What do parents want from schools? Evidence
from the internet. Educational Evaluation and Policy Analysis, 24(2): 133144.
19
Schneider, M., Marschall, M., Teske, P. & Roch, C. (1998a) School choice and
culture wars in the classroom: What different parents seek from education. .
Social Science Quarterly, 79: 489-501.
Schneider, M., Teske, P. & Marshall, M. (2000) Choosing schools: Consumer choice
and the quality of American schools. Princeton, NJ: Princeton University
Press.
Weekes-Bernard, D. (2007) School choice and ethnic segregation - educational
decision-making among black and minority ethnic parents. The Runnymede
Trust.
Weiher, G. R. & Tedin, K. L. (2002) Does choice lead to racially distinctive schools?
Charter schools and household preferences. Journal of Policy Analysis and
Management, 21(1): 79-92.
Wells, A. S. (1993) The sociology of school choice: Why some win and others lose in
the educational marketplace In E. Rassell and R. Rothstein (Eds.), School
choice: examining the evidence. Washington, DC: Economic Policy Institute. .
West, A. & Pennell, H. (1999) School admissions: Increasing equity, accountability
and transparency. British Journal of Educational Studies, 46: 188-200.
20
Figure 1: The percentage of main respondents giving none, one, two, or three
school choices on their LA common application form
21
Figure 2: The proportion of parents that get their first choice school (as
nominated on their LA application form), by the number of schools within a 3km
radius of their home
22
Table 1: Reasons for nominating the school named as 1st preference on the LA application form
Number of parents
% of parents that
Reason for school 'choice'
that give reason
give reason
School is near or nearest to home
6260
67.28
His/her friends go or were intending to go there
1837
22.43
His/her brother/sister went/go there
4320
46.93
Other relative/parent went/go there
1241
13.25
Wanted them to go to a different school to friend(s)
73
0.79
Wanted them to go to a different school to brothers/sisters
39
0.42
Wanted them to go to a different school to other relative/parents
31
0.25
How likely it was that he/she would get a place
556
6.38
School has good exam results/academic reputation
3814
43.55
General good impression of school
5483
62.87
School has strong anti-bullying policy
1105
12.52
School has small class sizes
1548
18.70
School caters for special needs
475
4.94
School offers specialised curriculum e.g. music, dance, activities
546
6.04
School has good facilities
3237
37.87
School offers childcare for parents who work or study
515
5.78
School is a feeder school
422
5.25
Religious grounds
978
10.24
Easy to get to on public transport
216
2.21
Ethnic mix of the school
470
3.91
School teaches in a language other than English
132
1.11
Wanted him/her to go to single-sex only school
1
0.01
Other reasons relating to the other children who go to the school
98
1.05
Did not like other/ local school
12
0.14
Reasons to do with nursery class/ pre-school
25
0.23
Connections (incl family) with school
45
0.51
Convenient location incl. close to childcare
123
1.42
Good school - good for specific reasons not elsewhere specified
4
0.02
Note: The table gives unweighted frequencies and percentages using survey weights. Parents were asked: “Which of these factors
were important in choosing to put this school as first choice on the form?” Multiple answers could be given. If their child was already
attending a nursery class at this school, parents were asked to give the reason they chose to send their child to that nursery class.
23
Table 2: Reasons for nominating the school named as 1st preference on the LA application
form, grouped by type of reason
Number of parents that
% of parents that give
Reason for school 'choice', grouped
give reason
reason
Proximity/ease of travel
6331
67.95
Sibling Rule
4320
46.93
Other family/friends
2874
33.00
Academic standards
3814
43.55
General good impression
5489
62.90
School characteristics and facilities
4512
52.36
Strategic
927
10.97
Religion
978
10.24
School composition
558
4.84
Pre-school & childcare
653
7.30
Note: The table gives unweighted frequencies and percentages using survey weights. Parents were asked: "Which of these
factors were important in choosing to put this school as first choice on the form?" Multiple reasons could be given If their
child was already attending a nursery class at this school, parents were asked to give the reason they chose to send their
child to that nursery class. Details for the way reasons were grouped can be found in appendix table 1.
24
Table 3: Most important reasons for nominating the school named as 1st preference on the LA
application form, grouped by type of reason
Number of parents that give
Reason for school 'choice', grouped
reason
% of parents that give reason
Proximity/ease of travel
2,566
25.56
Sibling Rule
2,350
24.98
Other family/friends
466
5.12
Academic standards
1,521
16.62
General good impression
1,439
16.72
School characteristics and facilities
475
5.19
Strategic
62
0.73
Religion
315
3.27
School composition
44
0.4
Pre-school & childcare
97
1.04
Note: The table gives unweighted frequencies and percentages using survey weights. Following the question on the
reasons for their choice of school, parents were asked: "And, which of these was the single most important reason (for
putting this school as first choice)?" Details for the way reasons were grouped can be found in appendix table 1.
25
Table 4: Most important reason for school choice, by the highest educational or vocational qualification of the main respondent
Highest educational or vocational qualification of the main respondent
Most important reason given
Missing
Proximity/ease of travel
Sibling Rule
Other family/friends
Academic standards
General good impression
School characteristics and facilities
Strategic
Religion
School composition
Pre-school & childcare
Total
None
0.88
40.64
29.07
7.00
7.80
8.95
3.69
0.15
1.58
0.11
0.12
100
'Other'
academic qual,
or vocational
0.05
31.99
26.16
4.75
13.29
13.94
4.95
0.32
3.81
0.35
0.39
100
GCSE
grades D-G
0.16
29.70
24.62
7.91
12.21
13.86
6.72
0.60
2.77
0.46
0.99
100
GCSE
grades A*-C
0.33
24.06
26.25
5.59
16.84
16.55
5.46
0.65
2.81
0.36
1.09
100
AS/A level or
diploma in
higher ed.
0.39
20.14
24.41
4.34
20.89
17.55
5.21
0.84
4.24
0.38
1.60
100
Degree +
0.41
20.42
21.06
3.08
20.27
22.78
4.56
1.37
4.32
0.59
1.16
100
Note: The table gives percentages using survey weights. The table should be read down the columns, for example 40.64% of those with no educational qualification give 'proximity' as
the most important reason for their choice of school.
26
Table 5: Most important reason for school choice, by the main respondent's socio-economic status (in quintiles)
Socio-economic status (SES) of the main respondent
Most important reason given
Missing
Proximity/ease of travel
Sibling Rule
Other family/friends
Academic standards
General good impression
School characteristics and facilities
Strategic
Religion
School composition
Pre-school & childcare
Total
Lowest
quintile
0.5
32.21
29.08
6.8
9.95
12.03
6.6
0.31
1.4
0.38
0.75
100
2nd quintile
0.39
27.67
25.65
6.05
15.43
14.64
5.45
0.18
3.1
0.43
1
100
3rd quintile
0.27
26.53
27.59
4.8
16.34
14.92
5.18
0.53
2.86
0.24
0.74
100
4th quintile
0.29
21.73
23.97
5.11
19.08
18.13
4.63
0.9
4.57
0.44
1.16
100
Highest
quintile
0.39
18.38
20.73
2.92
22.1
23.39
4.46
1.57
4.12
0.39
1.54
100
Note: The table gives percentages using survey weights. The table should be read down the columns, for example 32.21% of those with the lowest SES choose
'proximity' as the most important reason for their choice of school. The variable for SES was created using principal component analysis. Details of this procedure can
be found in appendix 2. The lowest quintile contains the fifth of the sample with the lowest socio-economic status.
27
Table 6: Mean characteristics of all schools in the feasible choice set, split by parent characteristics
Mean school characteristics
Proportion of students eligible for FSM
Proportion of students with EAL
Proportion of students with some SEN
Proportion of students that are white
Proportion of students with level 5 in all KS2 subjects
KS2 mean score (averaged over 3 years)
Proportion of schools: community schools
Proportion of schools: voluntary aided schools
Proportion of schools: voluntary controlled schools
Proportion of schools: foundation schools
"Catchment area"
Full
sample
0.16
0.11
0.19
0.87
0.19
27.79
0.56
0.33
0.09
0.01
2064.04
Parental characteristics
No
Low SES High SES education
0.22
0.11
0.23
0.12
0.08
0.18
0.21
0.17
0.21
0.84
0.90
0.77
0.17
0.23
0.16
27.34
28.28
27.25
0.58
0.55
0.55
0.35
0.32
0.38
0.06
0.11
0.06
0.01
0.02
0.01
1894.45
2251.61
1741.62
Degree +
0.12
0.11
0.18
0.87
0.22
28.22
0.54
0.33
0.11
0.02
2360.38
Note: Weighted values are shown. Statistics are calculated from all schools in the feasible choice set. The feasible choice set is defined as all
schools for which the pupil lives within the schools' catchment area, in the same LA and within 20km. Catchment areas are defined by the
straight line distance in which 80% of pupils in the previous cohort lived. Those with 'low SES' are those parents who are in the lowest socioeconomic quintile, 'high SES' in the highest socio-economic quintile. Socio-economic quintile was computed using principal component
analysis; details can be found in appendix 2. The proportion of schools refers to the proportion of non fee paying schools in the pupils' feasible
choice set.
28
Table 7: The rank of school chosen within the feasible choice set: ranked by KS2 mean
The rank of chosen school, where schools are sorted by their KS2
mean
Those with 8
Those with 10
Those with 5 schools
schools
schools
Mean rank
N
Mean rank
N
Mean rank
N
Lowest SES quintile
3.45
208
5.70
70
6.36
25
2nd SES quintile
3.06
199
5.28
69
5.71
29
3rd SES quintile
3.11
211
4.38
65
5.70
10
4th SES quintile
2.74
222
3.86
66
4.59
17
Highest SES quintile
2.69
174
4.05
70
3.82
12
Note: Schools are ranked from highest to lowest, so a rank of 1 indicates the most highly scoring school in the feasible
choice set. The feasible choice set is defined as the schools for which the pupil lives within the catchment area.
Catchment areas are defined by the straight line distance in which 80% of pupils in the previous cohort lived. SES refers
to the main respondent's socio-economic quintile, defined using principal component analysis (This process is defined in
appendix 2).
29
Table 8: The rank of school chosen within the feasible choice set: ranked by the proportion of
students with FSM
Lowest SES quintile
2nd SES quintile
3rd SES quintile
4th SES quintile
Highest SES quintile
The rank of chosen school, where schools are sorted by their
proportion of students with FSM
Those with 10
Those with 5 schools Those with 8 schools
schools
Mean rank
N
Mean rank
N
Mean rank
3.51
208
5.26
70
8.12
3.09
199
4.81
69
6.39
3.08
211
4.03
65
6.80
2.50
222
4.08
66
4.41
2.62
174
3.72
70
3.82
N
25
29
10
17
12
Note: Schools are ranked from lowest to highest, so a rank of 1 indicates the school with the lowest proportion of
students with FSM in the feasible choice set. The feasible choice set is defined as the schools for which the pupil lives
within the catchment area. Catchment areas are defined by the straight line distance in which 80% of pupils in the
previous cohort lived. SES refers to the main respondent's socio-economic quintile, defined using principal component
analysis (This process is defined in appendix 2).
30
Table 9: The rank of school chosen within the feasible choice set: ranked by the distance from
the pupil's home
Lowest SES quintile
2nd SES quintile
3rd SES quintile
4th SES quintile
Highest SES quintile
The rank of chosen school, where schools are sorted by their distance
to the pupil's home
Those with 10
Those with 5 schools Those with 8 schools
schools
Mean rank
N
Mean rank
N
Mean rank
N
2.47
204
3.15
78
3.53
40
2.25
215
3.33
95
4.05
40
2.06
218
3.40
96
3.88
17
2.23
248
3.50
97
5.13
31
2.10
218
3.08
110
4.25
21
Note: Schools are ranked from lowest to highest, so a rank of 1 indicates the closest school (calculated using straight
line distances) in the feasible choice set. The feasible choice set is defined as the schools for which the pupil lives
within the catchment area. Catchment areas are defined by the straight line distance in which 80% of pupils in the
previous cohort lived. SES refers to the main respondent's socio-economic quintile, defined using principal component
analysis (This process is defined in appendix 2).
31
Table 10: The proportion of pupils that get their first choice
school (as nominated on their LA application form), by the
rank of the school in terms of KS2 mean attainment.
All students
Lowest SES
Highest SES
quintile
quintile
Rank of
KS2mean
KS2mean
KS2mean
choice
1
0.91
0.8
0.91
2
0.93
0.93
0.95
3
0.93
0.93
0.93
4
0.95
0.94
0.94
5
0.97
1
1
Note: The table gives weighted percentages. Schools are ranked from highest
to lowest, so a rank of 1 indicates the most highly attaining school in the
feasible choice set.
32
Appendix 1: Grouping of survey responses
Group
Response
Proximity/ease of travel
School is near or nearest to home
Easy to get to on public transport
Sibling rule
His/her brother/sister went/go there
Other family/friends
His/her friends go or were intending to go there
Other relative/parent went/go there
Wanted them to go to a different school to friend(s)
Wanted them to go to a different school to brothers/sisters
Wanted them to go to a different school to other
relative/parents
Connections (including family) with school
Academic standards
School has good exam results/academic reputation
Good impression of the school
General good impression of school
Good school - good for specific reasons not elsewhere
specified
Did not like other/ local school
School characteristics/facilities
School has strong anti-bullying policy
School has small class sizes
School caters for special needs
School offers specialised curriculum e.g. music, dance,
activities
School has good facilities
Wanted him/her to go to single-sex only school
School teaches in a language other than English
Strategic
How likely it was that he/she would get a place
School is a feeder school
Religion
Religious grounds
School composition
Ethnic mix of the school
Other reasons relating to the other children who go to the
school
Pre-school &childcare
School offers childcare for parents who work or study
Reasons to do with nursery class/ pre-school
Convenient location incl. close to childcare
N
6260
216
4320
1837
1241
73
39
31
45
3814
5486
4
12
1105
1548
475
546
3237
1
132
556
422
978
470
98
515
25
123
33
Appendix 2: Detail of socio-economic status derivation
Socio-economic status: Principle Component Analysis
We create a single measure of parents’ socio-economic status (SES) for our
analysis instead of including many separate but correlated variables, following
the method used by the IFS in recent work. We use principal component
analysis (PCA) to identify the component which explains most of the variance
in SES. Based on this “1st component” (which explains 42% of the variance in
SES) we split the sample into quintiles. Those in the lowest SES quintile have
the lowest SES, and those in the highest SES quintile have the highest.
The variables we used in PCA are as follows:
1) Whether the main respondent reported financial difficulties. This is a
binary variable taken from wave 3 of the survey.
2) The highest National Statistics Socio-Economic Status (NSSEC) of the
household, taken from wave 3 of the survey.
3) Housing tenure, taken from wave 3 of the survey. Whether the
household lives in privately rented accommodation, social housing, or a
mortgaged/owned property are entered as binary variables.
4) Logged equivalised household income. We average measures of
household income over all waves, and weight for the number of people
and type of person in the household using the McClemens score.
Household members were given the following scores to weight by:
First adult
0.61
Spouse/partner
0.39
Other second adult
0.46
Third adult
0.42
Subsequent adults
0.36
Dependant aged 0-1
0.09
Dependant aged 2-4
0.18
Dependant aged 5-7
0.21
Dependant aged 8-10
0.23
Dependant aged 11-12
0.25
Dependant aged 13-15
0.27
Dependant aged 16+
0.36
34